THE COMPACT MUON SOLENOID (CMS) DETECTOR
PROPOSAL 892
UNITED KINGDOM
Bristol University; Brunel University; Imperial College, London; CCLRC Rutherford
*
*
Appleton Laboratory; Univ of the West of England ; Univ of Strathclyde
OTHER CERN MEMBER STATES
AUSTRIA: HEPHY, Vienna; BELGIUM: Univ Libre, Brussels; Vrije Univ, Brussels; Univ Cath.
Louvain, Louvain; Univ Antwerpen, Wilrijk; Univ Mons Hainaut, Mons; CERN; FINLAND: Helsinki
Inst. Phys., Helsinki; Phys. Dept., Helsinki Univ, Helsinki; Helsinki Univ Tech, Helsinki; Lappenranta
Univ Tech; Tampere Univ of Technology, Tampere; FRANCE: LAPP, Annecy; Inst. Phys. Nucl., Lyon;
LPNHE Ecole Polytechnique, Palaiseau; DAPNIA, Saclay; CRN-ULP, Strasbourg; GERMANY: I. Phys.
Inst., RWTH Aachen, Aachen; III. Phys. Inst. A, RWTH Aachen, Aachen; III. Phys. Inst. B, RWTH
Aachen, Aachen; Exp. Kernphysik, Univ Karlsruhe, Karlsruhe; GREECE: Univ Athens, Athens; NRCPS
Demokritos, Attiki; Univ Ioannina, Ioannina; HUNGARY: KFKI, Res. Inst. Part & Nucl. Phys.,
Budapest; Inst. Nucl. Res. ATOMKI, Debrecen; Kossuth Lajos Univ, Debrecen; ITALY: Dip. Fisica
dell’Univ and Sez. dell’INFN: Bari; Bologna; Catania; Firenze; Genova; INFN Milano; INFN Napoli;
Padova; Pavia; Perugia; Pisa; Roma I; Torino; POLAND: Inst. Experim. Physik, Univ Warsaw,
Warszawa; Soltan Inst. Nucl Studies, Warszawa; PORTUGAL: LIP, Lisbon; SLOVAK REPUBLIC:
Slovak Inst. Technology, Bratislava; SPAIN: CIEMAT, Madrid; Univ Autonoma, Madrid; Cantabria
Univ, Santander; Univ Oviedo; SWITZERLAND: Univ Basel, Basel; PSI, Paul Scherrer Institute,
Villigen; ETH, Eidgen. Techn. Hochschule, Zürich; Univ of Zurich, Zurich
NON-MEMBER STATES (Except USA)
ARMENIA: YerPhI, Yerevan; BELARUS: INP, Minsk; Byelorussian State Univ, Minsk; NCPHEP,
Minsk; RIAPP, Minsk; Dept. of Physics, Byelorussian State Univ, Minsk; BRAZIL: Univ do Estado, Rio
de Janeiro; Univ Federal, Rio de Janeiro; BULGARIA: Inst. Nucl. Res. + Nucl. Energy, Sofia; Univ of
Sofia, Sofia; CHINA PR: Institute of High Energy Physics, Beijing; Peking Univ Beijing; Univ of Science
& Technology of China, Hefei; CHINA (TAIWAN): National Central University, Chung-Li; National
Taiwan University (NTU), Taipei; CROATIA: Univ of Split; Tech Univ Split; CYPRUS: Univ of Cyprus,
Nicosia; ESTONIA: Inst. Chemical Physics and Biophysics, Tallinn; GEORGIA: Inst. Phys. Academy of
Science, Tbilisi; Inst. HEP, Tbilisi State Univ; INDIA: Bhabha Atomic Res. Centre, Tata Institute
(HECR), Tata Institute (EHEP), Mumbai; Panjab Univ, Chandigarh; Univ of Delhi, Delhi; IRAN: IPM,
Tehran; IRELAND: Univ College Dublin; KOREA: Cheju Nat. Univ, Jeju; Chungbuk Nat. Univ
Chongju; Kangwon Nat. Univ, Chunchon; Wonkwang Univ, Iksan; Chonnam Nat. Univ, Kwangju;
Dongshin Univ, Naju; Seonam Univ, Namwon; Kon-Kuk Univ, Seoul; Korea Univ, Seoul; Seoul Nat
Univ, Seoul; Seoul Nat. Univ Educ., Seoul; Sungkyunkwan Univ, Suwon; Kyungpook Nat. Univ,
Taegu; NEW ZEALAND: Univ Auckland; Univ Canterbury, Christchurch; PAKISTAN: Quaid-I-Azam
Univ, Islamabad; RUSSIA: JINR, Dubna; INR, Moscow; ITEP, Moscow; Lebedev Physical Institute,
Moscow; MSU, Moscow; IHEP, Protvino; Petersburg Nucl. Physics Inst., St. Petersburg; SERBIA:
Vinca Institute, Belgrade; TURKEY: Cukurova Univ, Adana; Middle East Tech. Univ, Ankara; Bogazici
University, Istanbul; UKRAINE: Kharkov State Univ; NSC-KIPT, Kharkov Inst. of Phys. and Tech.,
Kharkov; Inst. of Monocrystals, Kharkov; UZBEKISTAN: INR, Uzbekistan Acad. of Sciences, Tashkent
USA
Boston University; Caltech, Pasadena; Univ of California at Davis; Univ of California at Los Angeles;
Univ of California at Riverside; Univ of California at San Diego, La Jolla; Univ of California at Santa
Barbera; Carnegie Mellon Univ, Pittsburgh; Fairfield Univ, Fairfield; Fermilab, Batavia; Florida State
Univ, Tallahassee; Univ of Florida, Gainesville; Florida Inst Tech, Melbourne; Johns Hopkins Univ,
Baltimore; Univ of Illinois, Chicago; Univ of Iowa, Iowa City; Iowa State Univ; Univ of Kansas; Kansas
State Univ; Univ of Maryland, College Park; Massachusetts Institute of Technology, Cambridge; Univ
of Minnesota, Minneapolis; Univ of Mississippi; Univ of Nebraska-Lincoln, Lincoln; Northeastern
Univ, Boston; Northwestern Univ, Evanston; Univ of Notre Dame; Ohio State Univ, Columbus;
Princeton Univ, Princeton; Purdue Univ, West Lafayette; Rice Univ, Houston; Univ of Rochester; State
Univ New Jersey, Rutgers; Texas Tech Univ, Lubbock; Virginia Polytech. Inst. and State Univ,
Blacksburg; Univ of Wisconsin, Madison; Yale University, New Haven
INTRODUCTION AND OVERVIEW
The UK groups in CMS have lead responsibilities for two major CMS subsystems:
the Electromagnetic Calorimeter Endcaps, and the Tracker readout electronics, as
well as for the design and implementation of the Global Calorimeter Trigger system.
The UK also plays a central role in the development of software for reconstruction
and physics analysis, and in the development of GRID-based computing facilities.
The construction phase of the experiment follows an assembly sequence that allows
a complete detector (except for some elements of the Endcap Muon system) to be
ready for the first LHC beam in April 2007.
Civil Engineering and Assembly
Civil engineering works at Point 5 (located at Cessy, France) are advancing well.
The second phase of the construction of the surface hall has started with the removal
of the crown of the experiment shaft and the extension of the SX slab. The surface
control room building should be delivered to CMS in January 2005. Concreting of
the floor of the two underground caverns, UXC5 (for the experiment proper) and
USC5 (for the counting rooms and services) is advancing, and they are expected to
be given to CMS in summer 2004, when the installation of infrastructure will start.
The CMS experiment cavern will thus be ready to receive detector elements around
July 2005. Problems have been experienced with water ingress in two shafts, PM54
(services) and PX56 (experiment). While the ingress in PM54 is small and is not
considered critical, the problem in shaft PX56 has to be repaired. The repair is
foreseen straight after the delivery of UXC5/SX5 complex in summer 2004.
Magnet
All five barrel-yoke rings and six endcap-disks of the magnet yoke have been
assembled on the surface at Point 5. The outer vacuum tank is installed in the
central barrel-wheel, while the inner vacuum tank is currently in a vertical position,
wrapped in super-insulation, ready to be equipped with the inner thermal shield in
October 2003. Current leads have been manufactured and tested to 20kA at Saclay.
The power converter has been delivered and the power circuit will be tested starting
in December 2003. The dummy thermal load and the 6000 litre cryostat have been
installed on the central wheel and the outer cryogenics will be tested early next year.
All 21 lengths of the reinforced conductor (superconducting strands, pure Al insert
and Al alloy reinforcement) have been produced at Techmeta, France. The first two
coil modules have been completed and coupled together in Ansaldo (Genova, Italy).
The fifth and last coil module should be delivered to CERN in June 2004.
Tracker
All module components have entered production. A third of the sensors from
Hamamatsu have been delivered, and the rate of delivery is approaching that
required.
Difficulties encountered in packaging of electronic chips and manufacture of pitch
adaptors have been resolved and the production is now proceeding. All contracts
for opto-electronics have been signed. The procurement of a large part of the
mechanics is well advanced.
System tests of the tracker modules have been successfully carried out in May 2003
using the ‘25 ns bunched’ beam. A series of prototype FED modules have been
under test since the beginning of 2003. The manufacture of pre-series FEDs will
commence early in 2004.
The 0.25 µm deep-sub-micron (DSM) pixel readout chip (ROC) was received in
August 2003. The chip works well and minor corrections will be implemented in the
next submission. Pre-production modules using DSM ROCs will be made in 2004
and the mass production should start early in 2005.
Electromagnetic Calorimeter
About 21000 of the 62000 crystals for the ECAL barrel have been delivered and are
being used to construct modules containing 400 or 500 crystals in CERN and Rome.
Twenty nine modules have been produced out of 144. More than 120000 of the
140000 APDs have been delivered and the production is due to end in early 2004.
6900 VPTs have been delivered to RAL and most have been tested to 1.8T.
Mass production of the silicon sensors for the endcap preshower detector has started
in Russia and India (about 27% of the sensors have been produced and tested) and is
starting in Taiwan. Very encouraging performance has been attained with the DSM
chips (PACE3 and K-chip).
Hadron Calorimeter
Both HCAL half-barrels have been assembled at Point 5. Both of the endcapabsorbers have been mounted and optics (scintillators and fibres) installed. All of
the HO optics have been delivered to CERN.
All 36 steel absorber wedges for the HF have been produced and delivered to CERN.
The support structure for one side has been manufactured in Iran.
Muon Detector
About 90% of the 482 endcap cathode strip chambers have been assembled in sites in
China, Russia and the US. Roughly half have been tested with on-board electronics.
Over a quarter of the chambers are at CERN and after final tests about 20% have
been installed of the magnet yoke disks
CIEMAT, Aachen and Legnaro are continuing to produce barrel drift tubes at the
required rate of 18 chambers per year per site. Production at Torino is expected to
begin soon. Around 100 chambers (40% of the total) have been assembled and over
seventy of these have been delivered to CERN.
Production of the barrel RPCs is proceeding well. Over 100 chambers (20% of the
total) have been assembled and about 70 are at CERN. Two final RB1 chambers
have been operating in the gamma irradiation facility for over 6 months and so far
have integrated a charge corresponding to approximately the first 5 years of LHC
operation. Good performance has been observed.
Trigger and Data Acquisition System
Many full-function trigger card prototypes have been manufactured and final
validation tests are being done. Integration tests of detector primitive generators,
trigger system and DAQ are under way. Tests with a 25ns bunched ‘LHC-like’
beam have been very valuable to establish latencies. Software is now been
developed for testing and operation.
The Data Acquisition and High-Level Trigger Technical Design Report was
approved by the LHCC. Front End Readout Link (FRL) prototype boards are being
tested and preproduction has started. FED and Readout Builder (RB) demonstrators
are running at CERN. The first Event Filter sub-farm prototype is running under
realistic conditions.
TRACKER AND DATA ACQUISITION
UK Role
The CMS silicon tracker is a large silicon microstrip detector of about 220m2 area
arranged in a cylindrical geometry around the beam pipe, with petal-like module
arrangements in the endcaps. The system now contains almost 10 million
microstrips, each read out by an analogue front end chip (APV25) serving 128
channels, which samples and stores any signal present at the input every 25ns in a
pipeline memory for up to 4µs. Following a trigger, analogue pulse height data from
every detector channel are transferred via analogue optical links, with no zero
suppression, to an off-detector digitising module in an underground counting room
adjacent to the experimental cavern, approximately 100m distant from the tracker.
The CMS systems are designed to be able to cope with trigger rates of up to 100kHz.
The Tracker also contains a pixel detector inside the innermost microstrip layers but
there is no UK involvement in this part of the detector.
The Tracker contains approximately 15,000 silicon microstrip modules, comprising
mainly two sensors wire-bonded together, a front-end hybrid with readout and
service chips, carbon fibre frames and pitch adapters to match the sensor wire bond
pitch to the readout chip. To construct such a large number of modules to a uniform
quality in the relatively short time available, CMS adopted an industrial style
method of manufacture. Despite the widely dispersed community, a series of
identical manufacturiung centres were set up in a few central locations and these
centres are responsible for assembling, using automatic tooling, the modules from
components provided to them by other sites. There is a strong emphasis on
uniformity and quality control, but only recently has it been possible to demonstrate
production capability because all module components were not finalised. In
particular, the front-end hybrid took longer than anticipated.
The hardware deliverables for which the UK has major responsibilities are in the
electronic readout system. We are technically responsible for design and
development, and shared procurement, of the APV25 front-end readout chip and
other ancillary chips fabricated in the same 0.25µm CMOS process used throughout
the tracker, and the Front End Driver (FED) VMEbus readout board. We also have
shared responsibility for software required to operate and monitor the system, both
online and offline. In addition to various monitoring tasks we are delivering the
FEDs with the firmware and software required to integrate them into the CMS
system.
General tracker progress and project status
Over the last year, the tracker has begun to ramp up module production with the
objective of ~1000 modules per month by October 2003. A number of small
problems, each requiring attention to overcome, hampered progress initially but by
the summer it was believed that remaining problems had been addressed and
module production began in earnest, with great success. Production rates were still
limited by available components but it appeared that essentially all targets had been
met. Unfortunately in early September a new problem was encountered which
forced a temporary halt to further assembly until it was solved.
The multi-layer kapton-metal front end hybrid is mounted by lamination to a
substrate for mechanical support with an integral kapton flying lead and a miniature
connector. It was discovered that there was a weakness in the metallization of the
region close to the connector and many hybrids showed evidence of cracking even
during careful handling. The hybrid manufacturers and assembly company
responded rapidly and it was quickly identified that the origin of the problem was
in nickel plating of the copper prior to gold plating. It was established that this leads
to brittleness in the cable. The solution was to glue a FR4 “rigidifier” to prevent cable
flexing. The manufacturing process has been modified to laminate this piece to the
hybrids in future.
A second recent high priority issue is sensor production. The 300µm thick TIB (Inner
Barrel) sensors are manufactured by Hamamatsu, Japan. So far one third of the
production has been completed on schedule with very high quality deliveries and
there are no concerns about the remainder. The TOB (Outer Barrel) and TEC (Endcap) detectors are equipped with 500µm sensors produced by ST, Italy. In this case,
production was slow to ramp up and the quality of deliveries did not match CMS
requirements. It is important that a high production yield be achieved since
manpower is insufficient to re-test each sensor completely for the full production
period. Rejections are now mainly due to high leakage currents in many strips.
Several actions have been taken over the last year which raised quality significantly,
but the present acceptance rate by CMS is still too low, although many batches are
100% accepted. CMS is in discussion with ST to resolve this. The ultimate challenge
is to construct almost 16000 modules by the end of 2004 in an unavoidably large
number of variations (26 types with 14 sensor masks, 24 pitch adapters, 3 hybrid
layouts, etc.)
Cumulative module production
18000
16000
14000
Total
TOB
TIB TID
TEC
12000
10000
8000
6000
4000
2000
0
Oct-02
Jan-03
Apr-03
Jul-03
Oct-03
Jan-04
Apr-04
Jul-04
Oct-04
Figure 1. The target production rate of CMS tracker modules based on availability of
components and module assembly capacity.
Regarding electronics, all ancillary ASICs are available from final production wafers
and have been packaged. High yields were measured and there are no concerns
about availability. There has been further good progress with the UK part of the
project since the last report.
Front End Electronics
ASIC development
During 2002 and 2003, we experienced lower than expected yield on some lots of
APV25 wafers. In most cases yield was in excess of 50%, considered highly
acceptable for a large chip, and there was no evidence from any of our studies to
show long term weakness correlated to lower yield wafers. Nevertheless, since a
similar pattern of behaviour was observed on some other projects, intensive studies
were carried out for possible correlation between yield and measured chip
parameters. No obvious link was found.
During early 2003, IBM in Burlington, USA began intensive investigations of wafers
and individual die which we had identified as abnormal. Several Failure Analysis
techniques were used. It now is demonstrated that the explanation for the yield
behaviour is variable thickness of an Inter-Layer Dielectric (ILD) attributed to
unique features of large high energy physics chip designs which typically use much
more of certain metal layers than most designs processed by IBM. It was discovered
that the presence of the extra metal has an impact on some of the Chemical Metal
Polishing stages of the process which are essential to planarise wafers to ensure a
high degree of flatness, which is required for reliable lithography.
IBM processed wafers to study the effect in detail which we subsequently measured
in our usual way and thus optimised the ILD etching. Since then we have observed
the highest yields ever from APV25 wafers, with typical average lot yields exceeding
80% and individual wafers having yields of up to 95%. Deliveries of 48 wafers per
month are proceeding to complete APV25 production by April 2004. Imperial
College tests two wafers each day on most working days. This schedule is
considerably more aggressive than anticipated in early 2003 but is still not expected
to present difficulties.
FED developments
The last year saw a lot of progress with the Front End Driver. Eleven modules have
been assembled and until recently no major problems had been experienced. The
first FED was delivered to CERN in early November to integrate into our DAQ
system, then subsequently into large scale assembly test centres, of which the first
will be in Pisa in January 2004. This is good progress since the very first two FEDs
were delivered only in the third week of January 2003. Only a small number of
minor faults were found with the design and tests have been proceeding according
to plan.
In June a FED was equipped for the first time with all 8 optical receiver modules and
a 24-channel FED tester designed and manufactured at Imperial College was
brought into operation in July. In late August a first integration test was carried out
at Imperial, in which FED, FED tester, XDAQ software were fully operated together
for the first time.
A 96-channel tester exercises all FED channels simultaneously. It generates known
sequences of APV-emulated data with controllable amplitudes and patterns
allowing comparison of the data going into the FED with that emerging from it.
Because of the complexity of the optical inputs and need for precise temperature
control, it is only possible to fit 24 channels onto a single 9U board so 4 boards, now
complete, are needed to operate the FED. More detailed analogue and digital tests
are under way and evaluation of long term behaviour can be studied.
One problem identified during studies of the analogue performance of the FED, was
traced to the commercial ADC. An undershoot was visible before the pulse, which
was quite unexpected. Faults in the FED design and layout were eliminated and we
concluded the ADC was not behaving as it should, which was acknowledged by the
company after investigations. The ADC can be operated in two modes, with a 1V or
2V differential range. Fortunately, the postgraduate student who discovered and
measured this effect demonstrated that the 1V mode does not exhibit the
phenomenon. Since the FED is designed to use the ADC in the 2V mode, only a
small simple change is required. All other aspects of the analogue behaviour are
acceptable.
Figure 2.
modules.
The first assembled final Front End Driver containing all optical receiver
Until very recently the assembly quality had been without problems. Since no
problems had been experienced with five FEDs, 6 more were ordered. All six boards
were delivered in early October but had shorts between power and ground, traced
to solder flowing incorrectly under some of the FPGAs. The assembly company
identified contamination of the metal finish on the PCB as the origin of the fault and
observed that solder does not properly wet the metal. In fact, the metal finish on
these modules was changed from gold to tin, partly because of expected better long
term behaviour, but also because contamination of the gold on boards for other
projects. However, these manufacturing problems are not expected to prevent the
current schedule being met.
We are now planning the procurement of the ~500 FEDs needed by the tracker
system. This will take place in early 2005 following a pre-production order of ~20
FEDs to study in a fully populated crate.
The second major aspect of the FED development effort has been to provide
software to allow the FED to be installed in a common CMS system. This task has
being shared between Imperial and RAL very successfully.
Software
Offline software
UK institutes continue to make important contributions to the development of offline Tracker software. A UK physicist leads the Tracker `Data Handling’ group,
which is responsible for all Tracker offline test-beam analysis software, and for all
offline simulation and reconstruction software related to the Tracker readout and
data acquisition system.
This year saw the group releasing a flurry of CMS notes. Physicists from IC and
RAL participated in a test-beam analysis of the effect of Highly Ionizing Particles
(HIPs) on the Tracker performance. This analysis resulted in a change in the frontend hybrid electronics, so reducing inefficiency induced by HIPs to a very low level.
A Brunel physicist completed a study of the effect on the Tracker hit resolution of
the finite dynamic range of the Tracker FEDs. This confirmed that the current
dynamic range does not adversely affect the resolution. An ex-Imperial College
physicist wrote a note that compared the performance of various pedestal/noise
calibration algorithms.
More recently, Brunel and RAL have been collaborating with a physicist at CERN, in
attempts to run the CMS reconstruction program on the (computer) `filter unit’
during Tracker test-beams. This goal, which required writing a considerable
amount of new software, will bring us closer to the final CMS software environment,
since when LHC switches on, the reconstruction program running on the filter unit
will be responsible for high level trigger decisions and data quality monitoring.
Completing this task will be a major goal of the coming year.
Online software
The DAQ software for the tracker has been developed under XDAQ, which is a
distributed computing framework independent of hardware choice. XDAQ enables
the different aspects of the DAQ software to be loaded into any number of host
machines seamlessly. A simple system encapsulating the readout chain has been
developed, which is made up of Trigger Control; FEC; FED; Readout Unit; Event
Manager; Builder Unit; Filter Unit; Monitor Writer; Zebra Writer; Trigger Supervisor
( main control ). Each of these parts, whether software representations of hardware
or purely software, is represented by an instantiation of a C++ class.
These classes are then loaded into the XDAQ framework by the help of an XML
configuration file which describes the set-up, number of devices and settings for
each. Each of the hardware classes employ either the HAL (Hardware Access
Library) or a custom hardware access library written by L. Mirabito (Lyon), in order
to communicate with the VME memory space of the devices.
The FED part of this system has been developed at Imperial College while RAL
have been developing firmware (ID) and low level software (PPD). Integration of the
new FED classes developed between IC and RAL in a collaborative effort has been
very successful and the final goal is to provide a completely self contained software
package for the new FED which will seamlessly integrate with current DAQ systems
at CERN.
ELECTROMAGNETIC CALORIMETRY
Overview
The UK has the lead responsibility for the design and construction of the CMS
crystal Endcap Electromagnetic calorimeters (EE). This work is being carried out in
close collaboration with institutes from Russia together with groups providing
common items that will be used in both barrel and endcap regions of the detector.
The endcap calorimeters comprise a total of 14,648 slightly tapered lead tungstate
3
crystals, each approximately 3x3x22 cm in size, read out with one inch diameter
Vacuum Phototriodes (VPTs). The majority of the EE crystals are contained within
identical 5x5 units (25 channels) known as supercrystals (SCs). The crystals are
supported within the SCs by thin walled (400 mm) carbon fibre alveolar structures.
Current status
The CMS ECAL Endcap calorimeter has made steady progress in the areas of VPT
and mechanics procurement, in the design for the electronics integration on the Dee
backplates, and in the preparation for supercrystal and Dee assembly.
As of 31 Dec 2003, a total of 6900 vacuum phototriodes (500 pre-production, 6400
production) had been delivered to RAL, corresponding to 45% of the total. Nearly
6700 tubes have been tested to 1.8T at RAL and a sample of 733 tubes tested at 4T at
Brunel, as shown in Figure 3. Most of the devices showed excellent performance in
the acceptance tests.
The UK is responsible for the VPT high voltage system. Work is in progress to
design a cost effective cable, connector, and HV filter system from the counting
room to the VPTs. A number of possible cables and connectors have been appraised.
A final choice is expected as soon as the last sequence of high voltage tests has been
completed. This work is on the CMS critical path due to the need to pre-cable the
Endcap yokes in 2004. The orders for cables and fittings will need to be submitted
ahead of this exercise.
In 2003 a set of 100 EE pre-production crystals was received from the main crystal
producer in Bogoroditsk, Russia. 45 of these crystals have been measured for their
light yield properties in the ACCOS test facility at CERN and at the Crystal
Laboratory at Imperial. The crystals have good light collection uniformity. The
remaining 55 crystals have been measured at CERN and will also be appraised at
Imperial.
The EE major mechanics contract for 4 backplates, 4 ring flange supports to the
Hadron Endcap Calorimeter, and 600 positional spacers, was signed with a UK
company on 18 Aug 2003 for £450k following an international tender exercise. The
company has successfully produced a test model and has been given permission to
proceed with the rest of the contract. The mechanics for the first Endcap is due at
CERN on 14th May 2004.
The supercrystal mechanics contract is now complete, with only minor items such as
brackets yet to procure. Work continues to prepare the SC regional centre for
production.
There has been substantial progress in the design for the integration of the ECAL
electronics and cooling into the Endcap sections of the detector. The internal
routings of cables within the EE, to the EE patch panels, must be well understood by
the end of 2003. The cable exit positions from the EE patch panels will determine the
occupancy of the cable trays that pass over HE. This information will be crucial for
the pre-cabling of the CMS Endcap carts in 2004. An example of the integration
work is shown in Figure 4.
The mechanical aspects of electronics integration will be complemented by work on
E0′, a test bench of up to 4 supercrystals, to be eventually fitted with the new ECAL
electronics readout.
The TA1 group at CERN has started to prepare the Dee assembly lab in close
collaboration with Imperial College and RAL. An area has been secured in building
168 (adjacent to the UK liaison office). The TA1 group will manage the mechanical
assembly of Dees at CERN before shipping to building 867 for electronics
integration. The TA1 group has surveyed the OPAL Endcap frames for use on CMS.
The purchase of a loader arm to mount supercrystals onto Dees is under way in the
UK. A view of the assembly area with the frames is shown in Figure 5.
Figure 3. Delivery of Vacuum Phototriodes
Figure 4. An example of the detailed 3D design work on the back of a Dee.
Figure 5. Preparations for the Dee assembly area at CERN
ECAL electronics
A major review of the ECAL electronics was conducted by CMS in early 2002,
prompted by concerns about the technical status of the readout system and its cost.
Consequently, a modified architecture was proposed which substantially reduced
the number of optical links - one of the main cost drivers. The revised system makes
maximal use of developments from the CMS Tracker electronics, including the
control system and, especially, ASICs implemented in the IBM 0.25µm CMOS
technology. Imperial College and RAL engineers took on the design of several
critical elements. In particular they have developed a new digital chip, FENIX, and a
new amplifier in radiation-hard 0.25µm technology. Although this was a very late
stage for such a revision, the re-design proceeded extremely rapidly and is no longer
on the critical path of the overall ECAL project.
At the very front end of the revised readout system an amplifier senses signals from
the APD or VPT and the output is digitized with a 12-bit ADC. To cover the full ~16bit signal dynamic range multiple ranges are needed. This is accomplished by
providing three outputs with different gains. The existing commercial ADC chip
originally chosen was not matched to this and considerable cost savings could be
realised by a dedicated circuit. CERN developed a new multi-channel chip with 12bit ADCs in 0.25µm technology, with the aid of a specialist design house. This is
coupled to a multi-gain amplifier (MGPA), also in 0.25µm technology with 3
different gains, and selection of the appropriate one after the ADC. The digital logic
to do this is straightforward and is incorporated into the ADC chip. The first version
of the ADC did not reach the full design performance, but the causes are now
understood and will be corrected in the engineering run now in progress.
The amplifier was designed at Imperial College and laid out by the RAL
microelectronics group. The MGPA was submitted on the same MPW run as the
ADC and FENIX in February 2003. The production yield was high and the test
results are extremely good. It easily meets the stringent CMS linearity and noise
requirements. The energy equivalence of the noise, measured with a 50-channel
prototype of the electronics installed on an ECAL Super-module at CERN, averages
45 MeV (compared with the target of 50 MeV). Very minor improvements have been
implemented on a final version which is expected in Spring 2004. The MGPA has
another significant advantage compared to the old system: for the first time the
Endcap ECAL (VPT) readout can be implemented simply by altering two external
components on the front end board, whereas the previous system required a new
chip.
The FENIX chip follows the very front end to provide several basic functions; it is
configurable between them to minimize development cost and time and optimize
production and spares. These functions are storage of digitised raw data,
transmission of trigger primitives extracted from raw data to the L1 trigger cards,
and transmission of 10 samples of raw data for 5 crystals per event. Each front-end
card contains seven FENIX chips. The FENIX functionality is defined by VHDL
code, developed on an FPGA by groups at CERN and LLR, France. In parallel, the
design of the FENIX ASIC was undertaken by RAL ID. A board with an FPGA
emulating the FENIX chip was produced and operated in RAL and, subsequently, in
CERN. The FENIX chip design has been completed and fabricated on a CERN multiproject IBM run and is currently under test. The initial results show that it works
according to design. An engineering run incorporating some minor changes has
been submitted. The RAL ID group has also designed and delivered prototypes of
the associated Front-End printed circuit board.
GLOBAL CALORIMETER TRIGGER
The calorimeters provide two types of information to the Level-1 trigger:
electron/photon and jet candidates, collectively known as trigger ‘objects’; and sums
of transverse energy and its components over the whole region up to |η|=5. The job
of the Global Calorimeter Trigger (GCT) is to sort the different types of object,
passing on the best four of each type for use in the Level-1 decision, and to calculate
total and missing energy for the whole detector. These functions will be performed
using a single, generic design of trigger processor module making extensive use of
Field-Programmable Gate Array (FPGA) technology. The logic of the GCT is
illustrated schematically in Figure 6.
Figure 6. Schematic diagram of the CMS Global Calorimeter Trigger
The final processor design for the GCT has demanding requirements on data
input/output capabilities as well as logic performance. The input data of 5040 bits
per beam crossing will be received by eight processor modules, giving an input rate
of more than 25 Gbit/s per module. Recent developments in commercial
serialiser/deserialiser (serdes) chipsets allow us to achieve this density using readily
available, low cost components. Following an extensive programme of technology
evaluation in the UK, the prototyping of GCT components has begun during 2003.
The two main components of the system are the generic processor modules and the
input modules. A first set of prototypes of the input modules has been tested
extensively. During the summer some of these modules were taken to Vienna and to
Wisconsin and a successful series of tests of the interfaces to the Global Trigger and
Regional Calorimeter Trigger was undertaken. Figure 7 shows the setup for testing
the links to the Global Trigger, with a GCT prototype module sending data to a
Global Trigger receiver prototype via gigabit serial links.
Figure 7. Testing of the GCT input module prototypes and the interface to the Global
Trigger in Vienna, July 2003. The GCT components are on the right of the photograph.
The design of the processor module was completed during 2003. The module
contains 17 FPGAs from the Virtex-II family from Xilinx, including four large 3million gate devices to perform the algorithm processing. An aggregate data
input/output capacity in excess of 100 Gbit/s is provided by serdes devices at the
front and rear of the module. Testing of the first prototypes is now under way.
Production of the final GCT components is scheduled to take place in 2004 so that
the system can be installed, along with other central components of the trigger
system, during 2005.
PHYSICS AND COMPUTING
Physics Reconstruction and Selection
The PRS project
After completion of the work on the DAQ TDR in December 2002, which gave
detailed results on the complete selection chain from the Level-1 trigger through the
High Level Trigger, the four original ‘detector’ PRS groups – e/gamma, muon,
b/tau and jets/missing energy – were augmented by the addition of four new
‘analysis’ PRS groups: Higgs (led by a UK physicist), Standard Model, SUSY and
beyond the Standard Model, and Heavy Ions. The aim is to organize physics,
reconstruction and selection studies, including development of the necessary Object
Oriented software, towards the milestone of the Physics TDR, due at the end of
2005.
The Physics TDR is meant to be a test of the validity and readiness of software and
computing, as well as of the collaboration’s knowledge and skills. It will be
organized in two volumes. Volume I will cover detector response, physics objects,
calibration, alignment etc. This is the work of the detector PRS groups. Volume II
will cover high-level analysis and will be divided into two parts. The first part will
deal with a small number of very full analyses, to demonstrate how we can do
physics. This again is work for the detector PRS groups. The second part will cover
general physics topics, giving an overview of what physics can be done by CMS.
This is the part for which the new analysis PRS groups were created.
The broad aims of the work towards the Physics TDR can be summarized as follows:
1. Complete the physics reconstruction and selection software of CMS.
2. Complete the simulation software, obtain a detector parameterization, and build it
into a new fast simulation.
3. Engage the collaboration, activating and training the maximum possible number
of people.
4. Create a strategy for physics as a function of luminosity.
5. Put in place, ahead of time, all the recipes and methods that will be used to
commission CMS for physics.
ECAL – e/gamma Group
The ECAL-e/gamma group is responsible for all ECAL software and reconstruction
and physics-related tasks, including calibration, test-beam analysis and simulation.
A UK physicist has led the ECAL-e/gamma group since its inception, and the
ongoing UK contribution consists of one physicist and one student working fulltime, plus a number of additional part-time contributors.
Work this year has required the detailed definition of the data samples to be
simulated for the 2004 computing data challenge (DC04). These Monte Carlo
samples will also be used for physics studies for the Physics TDR. This has
necessitated seeking agreement of the many interested parties, and organizing the
definition of samples totalling some 20M events, which have now been generated
(mostly with PYTHIA), then simulated using both GEANT3 and later GEANT4. The
simulation of the digitization, the final step, will begin in 2004.
Much attention and work has been devoted to breaking down the work required to
build and complete the ECAL software into well defined packages that can be taken
on by groups and institutions. This effort culminated in a Tasks and Responsibilities
document that was presented and made available in the December CMS Week. One
of the challenges of this effort is the range of software covered: while it easy to
assign specific hardware related software to an institution, this is not an appropriate
model for physics studies, and the field encompassed by ECAL-e/gamma spans
these extremes.
A substantial contribution has been made in following and co-ordinating the studies
of many groups of physicists at many institutions.
One of the most important milestones in 2003 has been the replacement of the old
(Fortran based) GEANT-3 particle transport and detector simulation program, by its
Object Oriented GEANT-4 successor. This has required detailed verification of the
transported geometry model, and comparison of new results with old. UK physicists
have played an important role in this work. A UK physicist was responsible for the
implementation and testing of the ECAL geometry in the new CMS GEANT4,
OSCAR program.
UK students and physicists have also contributed to the test-beam work where a
new software framework for analysis has been developed this year, in preparation
for the testing and pre-calibration of complete super-modules in 2004. Energy
resolution, and the details of energy reconstruction from the 40MHz ADC time
samples of the new front end electronics, is being studied by UK students.
Work by UK physicists on the end cap trigger tower mapping has continued,
although final results have been delayed by serious delays in the Computing and
Core Software (CCS) implementation of the new (POOL based) persistency
mechanism which has now replaced the commercial Objectivity package. The
delays, which are widespread in both the CCS and PRS domains, result from serious
staffing shortages, which are widely acknowledged, and to alleviate which
considerable efforts are being made.
A UK student has started investigating some of the aspects of ECAL intercalibration. This is an area where a much more widespread and coordinated effort is
needed in the future.
After an ECAL geometry workshop in November, which highlighted the need for a
new maintainable OO-based geometry description, a UK physicist took on the
responsibility of providing the software for the Endcap geometry description.
Higgs group
The newly formed Higgs group is led by a UK physicist. A further UK physicist and
three UK students are working on Higgs decay channels, and on the related radion
channel, which also falls within the scope of the group. The radion study, “Search
for the radion decay into a Higgs boson pair with final states γγ+bb, ττ+bb and
bb+bb” was presented at a CMS plenary Physics meeting in the December CMS
week.
As in the case of the ECAL-e/gamma group, detailed Monte Carlo requests for the
Higgs group have been prepared, organized and submitted for production in
preparation for the DC04 data challenge. A UK physicist has produced an interface
for the dedicated leading order event generators ALPGEN, MadGraph, and
CompHep, so as to be able to produce event samples within the CMS production
environment.
A UK physicist was convenor of the Higgs group at the 2003 Les Houches
Workshop, and produced results for the workshop proceedings, which were
approved, on the subjects of radion → hh → γγ + bb channel, and the expected
+
precision of tan(β) measurement in MSSM H→ττ and H → τν channels.
A collaboration has been initiated with Southampton theorists to study the effects of
SUSY particles on the generation and decay of SUSY higgs particles. No
comprehensive study of these effects has previously been performed. The
collaboration will also encompass possible CP violation effects in the Higgs sector,
particularly in NMSSM models.
Other physics work
Jet reconstruction algorithms based on 'energy flow', making maximum use of
precision tracking and electromagnetic calorimetry, are likely to play a major role at
the LHC. A UK physicist has led one recent study of the use of these techniques in
CMS and the results of this study have been approved for outside presentation at a
recent CMS Physics Meeting
A UK physicist coordinates the Gauge Coupling subgroup of the CMS Standard
Model Physics Group. A further UK physicist and two students are active in this
group, investigating triple and quartic gauge couplings.
GRID computing for CMS
The CMS work plan for GridPP is organised into three distinct but overlapping
areas: workload management, data management and monitoring/metadata
management. The tools and approaches developed are exercised in the context of
large-scale data challenges, with a leading UK contribution.
CMS will shortly be undertaking its most ambitious data challenge to date. ‘DC04’
requires the month-long operation of a 25% scale worldwide computing system. In
the UK, we will execute tens of thousands of jobs to produce ~40TB of simulated
and reconstructed data at T1 and T2 sites. We have made extensive tests to store and
replicate this data using regional centre resources, during which we had to solve
problems associated with the Replica Catalogue service and took part in 'DC04 PreChallenge' production which made extensive use of the Tier 1 centre at RAL and
resources at Imperial College. Over 70 million fully simulated and digitised events
were produced and stored.
We are completing the implementation of a web portal for Monte Carlo production
and analysis that will enable CMS users to flexibly submit production and analysis
jobs via a web interface. This will lead to initial work on a Grid-enabled analysis
interface for CMS users.
After production-scale deployment and stress testing of the POOL object store
product as part of LCG, we have begun the task of integrating the POOL catalogue
component with CMS Grid data management, monitoring, and workload
management systems. A prototype Grid-aware event buffer and WAN data
streaming system is being implemented, facilitating prompt reconstruction and
calibration tasks at the CMS T0 and T1 centres. An interface is being developed
between the existing CMS data management system (SRB) and Grid middleware.
We will continue the integration of the CMS COBRA (COBRA is the realization of
the CMS OO Software Architecture for High Level Trigger, Simulation,
Reconstruction, and Analysis) framework with the Grid, via RGMA. We have
recently achieved a dramatic improvement in the number of messages that can be
handled using RGMA and work closely with the EDG development team to
continually improve the reliability and performance of this key middleware.
CONCLUSION AND SUMMARY
The CMS experiment is making steady progress towards the completion of the
detector in time for the LHC start-up in 2007. The projects with direct UK
involvement, the ECAL Endcaps, the Silicon Tracker, and the Global Calorimeter
Trigger have all taken major strides towards completion in the past twelve months.
The production of tracker modules has been ramped up in 2003 towards the longterm objective of 1000 modules/month. Significant successes have been achieved in
the production of the APV25 wafers and the Front-End Drivers (FEDs), for which
the UK tracker groups are responsible. The typical yield obtained from the APV25
wafers has been increased during the year from around 50% to over 80%, and
APV25 production is expected to be completed by April 2004. Fully-equipped FEDs
were tested successfully in 2003, with a number of problems being overcome, and
the procurement of the total of about 500 devices will go ahead early in 2005, with a
pre-production order in 2004. In addition, excellent progress has been made in the
development of FED software.
The ECAL Endcap project has also moved forward significantly. The delivery of
VPTs from the supplier in Russia is approaching 50% of the total requirement, and is
proceeding according to the agreed schedule. Work has begun on the design and
implementation of the VPT high-voltage system.
The procurement of the
supercrystal mechanics is almost complete, and 100 pre-production Endcap crystals
have been received, allowing supercrystal production to begin in 2004. There has
been a substantial revision of the ECAL readout electronic system, on a very short
timescale, to which UK physicists and engineers have made substantial
contributions by exploiting their experience of the 0.25µm CMOS technology gained
on the CMS Tracker. This will ensure that the system meets the demanding CMS
precision requirements at a much reduced cost.
Prototyping of components for the Global Calorimeter Trigger began during 2003,
with extensive, and successful, tests on the first prototype input modules. In
addition, the processor module design was completed during the year, and
prototype testing is in progress.
The UK groups continued to make substantial contributions to CMS physics and
software during 2003. A significant achievement, in which UK physicists played an
important role, was the replacement of the Fortran- and GEANT3-based detector
simulation program by a new simulation based on Geant4 and written in C++. UK
groups are extremely influential in CMS Grid computing activities, playing a major
role in the development of Grid software, as well as in Monte Carlo production and
Data Challenges.
UK physicists and students have a leading role in the newly-formed Higgs physics
working group, as well as in many other physics studies, and are well-placed to
make major contributions to the CMS Physics TDR, whose publication is due at the
end of 2005.
PUBLICATIONS
Journal papers
Intercalibration of the CMS electromagnetic calorimeter crystals in φ using symmetry of energy
deposition
C. Seez et al
J. Phys. G: Nucl. Part. Phys. 29 (2003) 1299-1326
The response to high magnetic fields of the vacuum phototriodes for the Compact Muon Solenoid
endcap electromagnetic calorimeter
K.W. Bell, R.M. Brown, D.J.A. Cockerill, P.S. Flower, P.R. Hobson, D.C. Imrie, B.W. Kennedy,
A.L. Lintern, O. Sharif, M. Sproston, J.H. Williams
Nucl Instr & Meth A504 (2003) 255-257
Precision fitting function for HPMT spectra
D. Britton
Nucl Instr & Meth A504 (2003) 298-300
Conference contributions
Requirements from experiments in Year 1
T. Virdee
Paper presented at LHC Performance Workshop, Chamonix XII, March 2003
The CMS tracking readout and front end driver testing
M. Pearson
Paper presented at IOP Particle Physics 2003, Durham, April 2003
Lead Tungstate crystals for the CMS Electromagnetic Calorimeter
M. Ryan
Paper presented at IOP Particle Physics 2003, Durham, April 2003
Highly Ionising Events in the CMS tracker
R. Bainbridge
Paper presented at IOP Particle Physics 2003, Durham, April 2003
The CMS trigger system
C. Seez
Paper presented at IV International Symposium on LHC Physics and Detectors,
FermiLab, May 2003
Higgs searches, Review of LHC potential
A. Nikitenko
Paper presented at Advanced Studies Institue – Physics at LHC, Prague, July 2003
Electroweak Physics and Top Quark Mass at the LHC
C. Mackay
Paper presented at EPS Conference on High Energy Physics, Aachen, July 2003
Where theoretical uncertainties may affect Higgs searches at LHC
A. Nikitenko
Paper presented at Workshop on Precise Cross-Section Measurements at the LHC, Binn,
Switzerland, October 2003
Early Physics Reach of CMS
A. Nikitenko
Paper presented at NSFC-CERN Workshop on LHC Physics, Weihai, China, October 2003
The CMS Trigger System
C. Seez
Paper presented at NSFC-CERN Workshop on LHC Physics, Weihai, China, October 2003
Scalability Tests of R-GMA Based Grid Job Monitoring System for CMS Monte Carlo Data
Production
H. Tallini, S. Traylen, S. Fisher, H. Nebrensky, P. Kyberd, D. Colling, P. R. Hobson, L. Field et al
Paper presented at IEEE Nuclear Science Symposium, Portland, OR, October 2003
Vacuum Phototriodes for the CMS Electromagnetic Calorimeter Endcap
K. W. Bell, R. M. Brown, D. J. A. Cockerill, P. S. Flower, P. R. Hobson, B. W. Kennedy, A. L. Lintern,
C. S. Selby, O. Sharif, M. Sproston, J. H. Williams
Paper presented at IEEE Nuclear Science Symposium, Portland, OR, October 2003
Evaluating the gamma and neutron radiation tolerance of Vacuum Phototriodes for the Endcap
Calorimeter of the Compact Muon Solenoid
P. R. Hobson, D. Cussans et al
Paper presented at the 8th ICATPP conference, Como, Italy, October 2003
Uniformisation of lead tungstate crystals for the CMS electromagnetic calorimeter endcaps
M. Ryan, D. Britton, X. Qu et al
th
Paper presented at the 8 ICATPP conference, Como, Italy, October 2003
The 96 channel FED Tester
G. Iles
th
Paper presented at 9 Workshop on Electronics for LHC, Amsterdam, October 2003
The CMS Tracker Front-End Driver
J. A. Coughlan, S.A. Baird, I. Church, C.P. Day, E.J. Freeman, W.J.F. Gannon, R.N.J. Halsall,
M. Pearson, G. Rogers, J. Salisbury, S. Taghavirad, I.R. Tomalin, E. Corrin, C. Foudas, J. Fulcher, G. Hall,
G. Iles, M. Noy, O. Zorba, I. Reid
th
Paper presented at 9 Workshop on Electronics for LHC, Amsterdam, October 2003
The MGPA ECAL readout chip for CMS
D.M. Raymond, G. Hall, J.P. Crooks, M.J. French
Paper presented at 9th Workshop on Electronics for LHC, Amsterdam, October 2003
Hardware and Firmware for the CMS Global Calorimeter Trigger
J. Brooke, D. Cussans, R. Frazier, G. Heath, D. Newbold, S. Nash, S. Galagadera, S. Madani, A. Shah
Paper presented at 9th Workshop on Electronics for LHC, Amsterdam, October 2003
LECC 2003 Closing remarks
G. Hall
Paper presented at 9th Workshop on Electronics for LHC, Amsterdam, October 2003
Refereed CMS Notes
Study of the Effects of Data Reduction Algorithms on Physics Reconsruction in the CMS ECAL
S. Rutherford
CMS NOTE 2003/001
Study of H→ττ with hadronic tau decays in CMS
A. Nikitenko et al
CMS NOTE 2003/006
Simulation of the HPMT/VPT Light Collection Ratio
D. Britton, M. Apollonio, E. McLeod
CMS NOTE 2003/011
CMS Test of the European DataGrid Testbed
D. Colling, B. MacEvoy, H. Tallini, O. Maroney, H. Nebrensky et al.
CMS NOTE 2003/014
Studies of PbWO4 Endcap Crystals Using a Hybrid Photomultiplier Tube
D.I. Britton, M.J. Ryan, X. Qu
CMS NOTE 2003/018
Test-Beam Analysis of the Effect of Highly Ionising Particles on the Silicon Strip Tracker
R. Bainbridge, I Tomalin et al.
CMS NOTE 2003/025
Calibration Studies for the CMS Microstrip Silicon Tracker on Simulated p-p Collisions
N. Marinelli
CMS NOTE-2003/021
Summary of the CMS Potential for Higgs Boson Discovery
A. Nikitenko, C.Seez at al
CMS NOTE 2003/033
CMS Internal notes
Results of Trigger Simulations with the Proposed ECAL Endcap Trigger Geometry
H. Heath, J. Brooke, G. Heath, D. Newbold, M. Probert
CMS IN-2003/009
A Simulation of the CMS ECAL 2002 M0-Prime Test Beam Using OSCAR
D. Holmes
CMS IN-2003/028
Defining ECAL readout units
C. Seez
CMS IN-2003/035
The Effect of the FED ADC Range on the Silicon Strip Tracker Coordinate Resolution
I. Reid
CMS –IN NOTE 2003/005
PhD Theses
Colour reconnection studies at LEP2 with the ALEPH detector, and data reduction algorithms for the
CMS electromagnetic calorimeter
S. Rutherford
Imperial College PhD thesis, 2003
Influence of Highly Ionising Events on the CMS APV25 Readout Chip
R. Bainbridge
Imperial College PhD thesis, November 2003
Light Collection Uniformity of Lead Tungstate Crystals for the CMS Electromagnetic Calorimeter
Endcaps
M. Ryan
Imperial College PhD thesis, December 2003